When Discovery's transporter inched the space shuttle to the launch pad in early April, NASA officials celebrated a significant milestone in efforts to recover from the Columbia disaster more than two years ago.
Now Discovery has been ordered back to the shed for more work. Its scheduled May 22 launch has been pushed to mid-July at the earliest. NASA wants to ensure that chunks of ice don't replace chunks of foam insulation as dangerous projectiles during launch.
Insulation from the external fuel tank broke loose during Columbia's climb into space in January 2003. It damaged the heat-resistant material on the leading edge of a wing. When the orbiter headed back to Earth at the end of the mission, it broke up during reentry some 16 minutes before touchdown, killing the seven-member crew.
This latest delay in Discovery's launch "is consistent with return to flight, not rush to flight," says Michael Griffin, who replaced Sean O'Keefe last month as head of the space agency.
The three remaining orbiters are slated for retirement in 2010 under President Bush's vision for space exploration. Until then, they are crucial to completing construction of the International Space Station. Discovery is carrying an Italian-made logistics module to the station.
As a result, "schedule does matter" in preparing orbiters for flight, Dr. Griffin said at a briefing on Friday. But efforts to meet schedules "shouldn't cause people to do dumb things."
During briefings on Friday, NASA officials explained that since the Columbia accident, engineers had focused largely on fixing the foam-insulation problem that triggered the tragedy. But they also conducted an exhaustive search for other sources of dangerous debris.
"We took 175 potential sources off our worry list," says Wayne Hale, the space shuttle program's deputy manager.
After reviewing records of past missions, loose ice emerged as a source of debris that needed attention. The combination of frigid liquid hydrogen and oxygen for the main engines, plus Florida's balmy, moist climate, allows ice to build on portions of the tank.
Some of that ice falls away in harmless directions. That is not the case, however, with ice from a 17-inch diameter feed line that supplies liquid oxygen to the engines. It runs along the outside of the tank. Dings to portions of the reusable solid-rocket boosters and on the orbiters' heat- resistant tiles have been traced to ice breaking free from fittings along this pipe. These fittings allow the line to expand and contract as the chilled oxidizer moves through it.
"We concluded we really had to do something about this ice," Mr. Hale says.
One approach might be to launch the shuttle in conditions that keep any ice somewhat loosely formed. But "what we're dealing with defies everyday intuition," he says, explaining that wind-tunnel tests showed that even loose agglomerations of ice can inflict significant damage.
The fittings have been known to release clumps of ice as large as five inches across, Hale notes. "A piece of ice that big traveling at three times the speed of sound can do some damage."
To deal with the problem, engineers will install heaters at the fittings along the feed line where ice tends to accumulate. But to refit Discovery's main fuel tank, the orbiter must be hauled back to the cavernous Vehicle Assembly Building.
Similar heaters are being installed on external tanks for missions that follow Discovery's return to flight.
At the time managers were deciding to roll Discovery to the pad, the wind-tunnel tests results weren't in yet, officials say. Managers felt they could provide the heaters on Discovery's next mission after return to flight as part of the normal upgrade process.
In many ways, Discovery's rollout and subsequent tests have been "a great learning experience for us," says William Parsons, the shuttle program's manager. The last time technicians had a shuttle on the pad for a preflight checkout was in October 2002.
One set of tests indicated that two of four key sensors were malfunctioning. All four are needed for an OK to launch. Discovery will stay on the pad while tests on those sensors continue.
The prospect of sending Discovery back to the shed "is a disappointment," Mr. Parsons says. "We'd like to have cleared these issues up" before rollout. "Now we'll take a step back and do it right."